Traveling reference spectroradiometer for routine quality assurance of spectral solar ultraviolet irradiance measurements
Post on 03-Oct-2016
Traveling reference spectroradiometer for routinequality assurance of spectral solar ultravioletirradiance measurements
Julian Grbner, Josef Schreder, Stelios Kazadzis, Alkiviadis F. Bais, Mario Blumthaler, Peter Grts,Rick Tax, Tapani Koskela, Gunther Seckmeyer, Ann R. Webb, and Diana Rembges
A transportable reference spectroradiometer for measuring spectral solar ultraviolet irradiance has beendeveloped and validated. The expanded uncertainty of solar irradiance measurements with this referencespectroradiometer, based on the described methodology, is 8.8% to 4.6%, depending on the wavelengthand the solar zenith angle. The accuracy of the spectroradiometer was validated by repeated site visitsto two European UV monitoring sites as well as by regular comparisons with the reference spectroradi-ometer of the European Reference Centre for UV radiation measurements in Ispra, Italy. The spectralsolar irradiance measurements of the Quality Assurance of Spectral Ultraviolet Measurements inEurope through the Development of a Transportable Unit (QASUME) spectroradiometer and these threespectroradiometers have agreed to better than 6% during the ten intercomparison campaigns held from2002 to 2004. If the differences in irradiance scales of as much as 2% are taken into account, the agreementis of the order of 4% over the wavelength range of 300400 nm. 2005 Optical Society of America
OCIS codes: 120.4640, 120.3940, 120.4140, 120.5630, 120.6200, 260.7190.
During the past decade a large number of monitoringstations have been established worldwide for moni-toring the spectrum of solar UV radiation reaching
the Earths surface. UV monitoring is considered oneof the most important activities that have been stim-ulated in past years by the observed decreases instratospheric ozone.1,2 Moreover, the association ofsolar UV radiation with damage to human beings andto the ecosystem in general and its strong relation toatmospheric chemistry imposed the necessity of per-forming high-quality spectral UV measurements thatwould help substantially to address these issues.3Given that the UV represents only a small part of thesolar spectrum, its measurement becomes difficult,requiring high-level technology as well as sophisti-cated instrumentation and procedures.4
The need for quality control (QC) and quality as-surance (QA) of UV measurements has been recog-nized since the beginning of the 1990s.5,6 Theestablishment of international databases of solar UVmeasurements, e.g., the European UV Database(EUVDB) established within the framework of Euro-pean Commission-funded projects and the WorldOzone and UV Database (WOUDC) hosted by theMeteorological Service of Canada, and in particular
When this research was performed, J. Grbner (email@example.com) was with the European Reference Centrefor Ultraviolet Radiation Measurements, Institute for Health andConsumer Protection, European Commission, Joint Research Cen-tre, Ispra, Italy; he is now with the Physikalisch-MeteorologischesObservatorium Davos, World Radiation Center, Dorfstrasse 33,CH-7260 Davos Dorf, Switzerland. J. Schreder is with CalibrationMeasurement Software Solutions, Kirchbichl, Austria, S. Kaza-dzis and A. F. Bais are with the Laboratory of Atmospheric Phys-ics, Aristotle University of Thessaloniki, Thessaloniki, Greece, M.Blumthaler is with the Institute for Medical Physics, University ofInnsbruck, Innsbruck, Austria, P. Grts and R. Tax are with theLaboratory of Radiation Research, National Institute of PublicHealth and the Environment, Bilthoven, The Netherlands, T. Ko-skela is with the Finnish Meteorological Institute, Ozone and UVResearch, Helsinki, Finland, G. Seckmeyer is with the Institutefor Meteorology and Climatology, University of Hannover, Han-nover, Germany, A. R. Webb is with the School of Earth Atmo-spheric and Environmental Sciences, University of Manchester,Manchester, England, and D. Rembges is with the Institute forHealth and Consumer Protection, European Commission, JointResearch Centre, Ispra, Italy.
Received 10 January 2005; revised manuscript received 18
March 2005; accepted 18 March 2005.0003-6935/05/255321-11$15.00/0 2005 Optical Society of America
1 September 2005 Vol. 44, No. 25 APPLIED OPTICS 5321
their relationship to the users community call forstrict application of QCQA procedures to ensurethe quality and comparability of the data,7 havehelped to fill that need. QC is performed at moni-toring stations through the development and appli-cation of appropriate procedures, most of whichhave already been tested and verified through in-ternational collaborations among UV instrumentoperators. It is however, uncertain, at how many ofthe existing UV stations proper QC is maintained.Until now, QA has been achievedwith particularsuccessmainly through participation of instru-ments in intercomparison campaigns.8 As the num-ber of deployed instruments is constantly increasing,such campaigns have become impracticable; in addi-tion, there is a risk of damaging the instruments oraltering their optical characteristics during transpor-tation and the interruption of their regular recordsfor long intervals.
The European Commission-funded project QualityAssurance of Spectral Ultraviolet Measurements inEurope through the Development of a TransportableUnit (QASUME) was launched in December 2001(Ref. 9; see http:lap.physics.auth.grqasume). Itaims at providing QA to spectral solar UV measure-ments conducted now by spectroradiometers operat-ing in Europe by establishing a reliable transportablespectroradiometer system that can be transported toany UV monitoring site in Europe and provide collo-cated measurements with the local site instrument.This on-site QA exercise should be viewed as an al-ternative to the intercomparisons performed previ-ously, in which spectroradiometers from differentparts of Europe were gathered at one location to per-mit their performance to be assessed during simul-taneous measurements.10 The advantages of theproposed approach are that local monitoring instru-ments do not need to be transported and are used intheir natural environment during the intercompari-son; furthermore, a site can be visited at regular in-tervals for checks on its stability over extended timeperiods. While this procedure offers a more realisticevaluation of a monitoring site, it places strict criteriaon the performance and operation of the travelinginstrument, which must be proved to be stable at alevel against which all other instruments will bejudged.
The QASUME traveling unit is composed of thespectroradiometer, its calibrating unit, an angularresponse unit, and a heliumcadmium laser. The lasttwo items are provided to the local site operator fordetermining the angular response of the sites detec-tor and the slit function of the spectroradiometer,respectively. In this paper, only the reference spec-troradiometer and its associated calibrating unit arediscussed. The home site of the QASUME travelingunit is the European Reference Centre for ultravioletradiation measurements (ECUV) at the Joint Re-search Centre of the European Commission at Ispra,Italy.
The spectroradiometer consists of a commerciallyavailable Bentham DM-150 double monochromatorwith a focal length of 150 mmmonochromator andwith 2400 linesmm gratings. The wavelength rangeis 250500 nm, and the entrance and exit slit widthwas chosen to yield a nearly triangular slit functionwith a full width at half-maximum resolution of0.8 nm.11 The smallest wavelength increment is0.0025 nm. The spectroradiometer has two entranceports, which can be selected by a remotely controlledinternal mirror. The solar irradiance is sampledthrough a specially designed entrance optic (CMS-Schreder, Model UV-J1002) which is connected to oneport of the spectroradiometer through a quartz fiber.The second entrance port holds a pencil ray mercurylamp (Oriel, Model 6035) which is used to check thewavelength setting of the spectroradiometer. UntilSeptember 2003 a side-window-type photomultiplier(PMT) was used as a detector; then it was replacedwith an end-window-type bialkali PMT (electrontubes 9250QB). The photocurrent is measured with asix-decade current amplifier, integrated for a 100 mstime window, digitized, and transferred to a com-puter for further data treatment and storage.
Because the instrument is designed for outdoor so-lar measurements, the whole spectroradiometer sys-tem including the data-acquisition electronics iscontained in a temperature-controlled box that is sta-bilized to a predetermined temperature with a preci-sion of 0.5 K.
Initially, the spectroradiometer was characterizedin the laboratory; the results pertaining to the mostimportant parameters are discussed below.
A. Wavelength Scale
The wavelength scale of the spectroradiometer wasinitially determined by use of spectral emission linesfrom mercury, cadmium, and zinc spectral dischargelamps. We obtained the relationship between thegrating angle and the wavelength by simultaneouslyminimizing the residuals at all measured spectrallines. The best result is obtained with a second-orderpolynomial with the resultant residuals all below0.02 nm. The stability of the wavelength scale is mon-itored with the pencil ray mercury lamp mentionedabove. Before every solar measurement, a fast scanthrough the 289.9 nm spectral line is used to checkthe wavelength alignment of the spectroradiometer.The wavelength repeatability, based on these mea-surements, is usually better than 0.01 nm during oneday of continuous measurements.12 However, the re-spective wavelength scales of the two entrance portswere found to differ between successive site visits, soa different method was required for checking thewavelength scale of the measured solar spectra. Theselected method uses a validated extraterrestrialspectrum13 in the wavelength range of 340350 nm todetermine the wavelength offset between the two.Then, to reduce possible errors of the algorithm in-duced by moving clouds, for example, we use a daily
5322 APPLIED OPTICS Vol. 44, No. 25 1 September 2005
average wavelength offset to adjust all measured so-lar spectra for this day. The resultant spectral wave-length shifts, tested with the SHICRivm tool,14 areless than 0.03 nm over the wavelength range of310500 nm.
B. Solar Irradiance Scale
One of the more difficult tasks in absolute solar spec-troradiometry is to reliably transfer the laboratoryirradiance scale, based usually on measurements ofcalibrated 1000 W quartz-halogen lamps, to outdoormeasurements of solar radiation. The two main prob-lems in that respect are, first, the large intensitygradient of the solar UV radiation, which covers morethan 6 orders of magnitude from 290 to 400 nm, and,second, the practical aspects of calibrating a spectro-radiometer in the field and ensuring its stability overdays and sometimes weeks.
1. LinearityWe investigated the first problem by determining thelinearity of the detector and the acquisition electron-ics. This was done in the laboratory with a 250 Wquartz-halogen lamp and by measurement of its ra-diation at various distances with the QASUME spec-troradiometer. The measurements discussed herewere obtained with the end-window PMT in use sinceSeptember 2003. To cover the whole intensity scaleencountered during solar measurements, additionalmeasurements were made with the Teflon diffuserremoved, which increased the measured signal by afactor of 100. The linearity of the spectroradiometerwas then obtained by a suitable combination of thesemeasurements. As can be seen from Fig. 1, the spec-troradiometer behaves linearly for photocurrents upto 3000 nA; at higher photocurrents, from 3000 to50,000 nA, the system exhibits a gradual nonlinear-ity of as much as 2%, which we compensate for byapplying a suitable correction to the measured pho-tocurrents.
The sensitivity of the side-window-type PMT in useuntil July 2003, however, exhibited a clear correla-tion with the received radiation dose, and thisprompted the change in 2003. Indeed, during amonth-long measurement period in spring 2003, itssensitivity decreased by 10%, of which it recovered8% after a three-week shutdown.15 Similarly, duringfield campaigns in 2003, variations in spectral sensi-tivity of the order of 34% were observed during rou-tine calibrations of the QASUME spectroradiometer.Although this effect was not observed in 2002, it wastaken into account in subsequent campaigns in 2003by calibration of the spectroradiometer between solarmeasurements as many as 12 times per day.
2. ECUV Irradiance ScaleThe irradiance scale of the QASUME spectroradiom-eter is based on a number of 1000 W free-electron-laser-type tungsten-halogen lamps traceable to theprimary radiation standard maintained at thePhysikalisch-Technische Bundesanstalt (PTB) inBraunschweig, Germany. Since 2002 one or two newirradiance transfer standards have been purchasedfrom the PTB each year and added to the referenceset of transfer standards held at the ECUV. Thisensemble is carefully monitored at regular time in-tervals and establishes the irradiance reference towhich the QASUME spectroradiometer is directlyreferenced. Figure 2 shows such a reference lampintercomparison performed in the ECUV laboratoryin February 2004. The variability of the lamps rela-tive to one another is of the order of 0.5%, well withinthe expanded uncertainties of 3% stated in the lampcertificates. It furthermore shows that an irradiancescale based only on a subset of the lamps, such as wasused in the past, did not introduce any significantbias.
One should note that irradiance transfer standards
Fig. 1. Laboratory determination of the linearity of the QASUMEunit fitted with an end window photomultiplier. The system uses asix-decade current amplifier; decades are denoted V1V6.
Fig. 2. ECUV Irradiance scale determined from the average offive radiation sources traceable to the PTB, Germany. The sub-group composed of the two radiation sources labeled F324 andF330 was used until the end of 2003. The radiation sources aretungsten-halogen filament lamps manufactured by Osram-Sylvania.
1 September 2005 Vol. 44, No. 25 APPLIED OPTICS 5323
delivered by different national standards laborato-ries can differ by several percent.16,17 A comparison inMay 2004 between the ECUV irradiance scale andthe irradiance scale of the Central UV CalibrationFacility (CUCF), traceable to the National Instituteof Standards and Technology, showed that the latterwas higher by 12% from 300 to 450 nm.18 In con-trast, measurements made with a transfer standardfrom the Helsinki University of Technology werewithin 1% of the average of the transfer standards ofthe ECUV.
3. Portable Irradiance ScaleBecause the QASUME spectroradiometer was de-signed to measure at locations far from its laboratory,a portable irradiance scale was devised. It is com-posed of a portable lamp enclosure19 (called a calibra-tor from now on), a set of 100 and 250 W tungsten-halogen lamps, and a c...